Rumor mill: AMD CEO Dr. Lisa Su confirmed earlier this year that Zen 5-based CPUs for both client and server applications would launch in the second half of 2024. A series of recent leaks have already given some clues about the new core architecture, and a new rumor now suggests that it could represent a massive upgrade over Zen 4.
This latest rumor comes from noted hardware leaker @Kepler_L2, who says that Zen 5 cores will be substantially faster than their Zen 4 counterparts. According to his post on the AnandTech forum, SPEC benchmark tests suggest that Zen 5 cores could be more than 40 percent faster than Zen 4, possibly making the next-gen AMD CPUs a solid upgrade over the current-gen ones.
It is worth noting here that the tipster didn’t say whether the figure is based on integer or floating-point metrics. The former would obviously be more useful as it would indicate significantly-improved performance in both single-threaded and multi-threaded workloads. There’s no word either on whether the quoted improvement is an average figure or if it’s “up to” 40 percent higher. Either way, it’s still just a rumor at this stage, so take it with a pinch of salt.
The Zen 5 core architecture is expected to power a range of CPU lineups over the next couple of years, including the Granite Ridge desktop processors that are tipped to hit the market later in 2024 under the Ryzen 9000 moniker. Other expected chip lineups based on Zen 5 are the Strix Point APUs that are also tipped to launch this year, as well as the Fire Range laptop CPUs and Strix Halo flagship APUs that are rumored to be released in 2025.
Finally, the company will also use Zen 5 “Nirvana” and 5C “Prometheus” cores in its next-gen EPYC server processors slated to hit the market later in 2024. According to earlier reports, these chips will be fabricated on TSMC’s 3nm process and will enter mass production in Q3 2024.
AMD has officially stated that Zen 5 will offer several upgrades over Zen 4. These include enhanced performance, improved efficiency, a re-pipelined front end, as well as integrated AI and machine learning optimizations.
